s
.T75
Thirteenth Series, No. 12 May 27, 1922
S^arlj^ra Ololbg? lulblm
Supplementing the Hillegas Scale
A Description of the Derivation and Use of the Nassau
County Supplement to the Hillegas Composition Scale
By M. R. TRABUE, Ph.D.
PROFESSOR OF EDUCATION, UNIVERSITY OF NORTH CAROLINA
Published by
3(rart|ar0 (Ball^B^. (Halambm llnitirrBUg
525 Weet 1 20th Street
New York Qty
M«
"oe'-aaft
Qlrarljprs CdaUrgf ?inlUtin
Thirteenth Series, No. 12 May 27, 1922
Published twelve times a year: Fortnightly in September, October, and May, and
monthly from November to April, inclusive. Entered as second class matter January
15, 1910, at the Post Office at New York, New York, under the Act of August 24, 1912.
Acceptance for mailing at special rate of postage provided for in Section 1103,
Act of October 3, 191 7, authorized.
Supplementing the Hillegas Scale
A Description of the Derivation and Use of the Nassau
County Supplement to the Hillegas Composition Scale
By M. R. TRABUE, Ph.D.
PROFESSOR OF EDUCATION, UNIVERSITY OF NORTH CAROLINA
reprinted from
Teachers College Record, Vol. i8, No. i, January 1917
revised edition
Published by
STparlirrH (EoUrgp, ©nlumbia llniuprHity
525 West 1 20th Street
New York City
OCT 23 '22
^C1A683863
^0 I
SUPPLEMENTING THE HILLEGAS SCALE*
During the spring session of 1916, some of the advanced stu-
dents in the department of educational administration at Teach-
ers College were engaged in an educational survey of Nassau
County, Long Island. > In the course of this survey the educa-
tional achievements of public school pupils were measured by
means of standard educational scales and tests. The measure-
ments of the quality of English compositions written in Nassau
County suggested very strongly the need for two supplements
to the Hillegas Scale.' In the first place, there seemed to be
a distinct need for a supplementary scale composed of composi-
tions of the same general type as those written by Nassau County
pupils, and, in the second place, it appeared desirable to set up
some tentative standards indicating the quality of English
compositions to be expected from the pupils of any given school
grade. The discussion which follows is a report of certain efforts
to supply these two supplements to the Hillegas Scale.
In spite of all the criticisms of and objections to the Hillegas
Scale, the fact remains that it is one of the most useful measur-
ing instruments in the whole field of education. The objective
character of the scale has been extremely useful. By means of
this scale it is now possible for one person to understand and
to identify the quality of written English indicated by another
person. Intelligent use of the Hillegas Scale by teachers and
supervisors tends to place emphasis on the quality of the written
composition rather than upon the personal relations of the
writer and the teacher. Any supervisor who employs the scale
*Copyright, 19 17, by Teachers College.
^ Trabue, M. R., Report of a Survey of Public Education in Nassau County,
New York. Albany: University of the State of New York, Bulletin No. 652,
December i, 191 7.
' Hillegas, M. B., A Scale for the Measurement of Quality in English Composition.
Published by the Bureau of Publications, Teachers College, New York City.
2 TEACHERS COLLEGE BULLETIN
with a group of his English teachers, and who then takes the
trouble to discuss with such teachers in the light of this experi-
ence just what constitutes general merit in English composi-
tion writing, will be able to show as a result a much greater
agreement of opinions and of school marks in this subject. In
other words, the use of the Hillegas Scale calls attention to and
defines fairly well what we mean by general merit in English
composition.
One of the defects which is quite frequently pointed out by
those who object to the Hillegas Scale is the fact that the first
three samples of this scale are not the compositions of children,
but are artificially constructed samples. To some of the other
samples the objection is made that they are so short as to afford
very little evidence of their true quality. As an example we
may point out Sample 534, which represents Quality 5.85 on
the scale.
Fluellen
The passages given show the following characteristic of Fluellen:
his inclination to brag, his professed knowledge of History, his com-
plaining character, his great patriotism, pride of his leader, admired
honesty, revengeful, love of fun and punishment of those who
deserve it.
This confused sentence is hardly sufficient to give one a definite
impression as to its quality. Objection has also been made
to the fact that the compositions appearing on the scale are not
all of the same type. Some teachers claim that they find it
difficult to compare the quality of two such compositions as
the one shown above and the one which follows it in the Hille-
gas Scale (Sample 196, Quality 6.75).
Ichabod Crane
Ichabod Crane was a schoolmaster in a place called Sleepy Hollow.
He was tall and slim with broad shoulders, long arms that dangled
far below his coat sleeves. His feet looked as if they might easily
have been used for shovels. His nose was long and his entire frame
was most loosely hung to-gether.
Some slight additional criticism has been based upon the in-
equality of the intervals between the values of samples on the
scale. The interval between the values of the first and second
samples is, for example, 1.83 units, while the interval between
SUPPLEMENTING THE HILLEGAS SCALE 3
the second and third samples is only .77 of a unit, and the inter-
val between the eighth and ninth samples is only .66 of a unit.
Many of the objections made to the Hillegas Scale are of very
little consequence other than that they offer excuses for not
making use of the scale. It is nevertheless worth while to at-
NASSAU COUNTY SUPPLEMENT TO THE HILLEGAS SCALE FOR
MEASURING THE QUALITY OF ENGLISH COMPOSITIONS
Directions for measuring: Compare the quality of your composition with the quality
of the samples on the scale. Assign to your composition the numerical value of that
evaluated sample which most nearly equals it in merit.
0.
.004
1.1
1.06
1.9
1.93
2.8
2.81
3.8
3.84
What I should like to do next Saturday
I went going on to the Dox Saturdaye dnd day we the boys and
I well going home and I well going the boys, and I will going these
read in and they to night, and we or night. I well going a ground
shalt and I gone out I will going to shea shouse and I will shoe or
the skill of the shea of night.
I intend to mak a snou man and make an fort and fort snou ball
at chidern and hau I whist ma frant carolyn cole what were me I
will going to the mauiss on Saturday.
Georga will come went me.
at night I will going out went my mother to the marce
I will mak the snou man and the fort in the moning and in the
aftermoon I will go to the mauies.
I whist there whest school on Saturday
one next S aturday I expect to go to the city leve next G aturday
to see my ofriend archie king I am going to grow to the baning
balys circus with hime next S aturday fefore I go I have to do my
jobs feedsing the cows ard horse ard chinkens and geese next Saturday
My friend is a very good fellow to go and see So my mother S iad
" If I do my work during Easter week vacation I can go to the barning
baley circus with, hime
Once a pon a time there was a girl. One day she asked me what
I was going to do next Saterday so I said, "I am going to go for a
swim." And she said, "thats
just where I am going to." next Saterday came we both went
down together. We came home at noon time, after dinner we
went to the picktures. There we had a good time. And then came
home at night.
I would like to go out in the after noon and play catching the
ball. Go over to Bertha's house and have a few girls to come with
me and be on each others side. I have a tennis ball too play with.
The game is that one person should stand quite aways from another
person and throw the ball too one then another. Someone has to
be in the middle and try too get the ball a way from someone then
she takes this persons place who she caught the ball from. Then
till every person has a chance.
TEACHERS COLLEGE BULLETIN
5.0
4.97
7.2
8.0
Next Saturday I should like to go away and have a good time on
a farm. I should like to watch the men plowing the fields and
panting corn, wheat, and oats and other things planted on farms.
Next Saturday I will go to the Pioneer meeting if nothing happens
so that I cannot go. I should like to go swimming but it is not
warm enough and I would catch a bad cold. I should like to go to
my aunts and drive the horses, I do not drive without some older
person with me, so I cannot go very often.
I should like to see my aunts cat and her kittens, too. I think
I can, to.
I should like to join my girl friends, who are :?oing to the city on the
9:05 A. M. train. They are going shopping in the morning and will
^Q have lunch to-gether, then they are going to the Hippodrome After
the Hippodrome, they are all going home to dinner to one of the
girls houses, she lives on Riverside Drive so they expect to take the
"Fifth Avenue Bus" up there. The evening will be devoted to
6 01 P'^ying games, .'dinging and dancing.
If I had a thousand dollars to spend, I think I would take a trip
to San Francisco by train with the rest of the family, and stop at
a sea-side hotel. It would be glorious to see the surf again, and
to escape from the cold blustering weather of December for the
balmy breezes of the ocean, and the whiff of orange blossoms.
We could take long drives under shady trees, visit the orange and
olive groves and bathe in the surf. Think of bathing in the ocean
n December!
Coming home again I should enjoy stopping at Yellow Stone
Park. It would be lots of fun to camp out, and to ride over the
prairies on frisky ponies. It would be very interesting to notice
the change of climate as we got farther east, and to go to bed on
the train one evening feeling warm, and waking up the next morn-
ing feeling very chilly.
I am afraid by the time I would get home a thousand dollars
7 22 would be pretty well used up; but if not I would like to give a party.
One Sunday, towards the end of my summer vacation, I was in
bathing at the Parkway Baths. In the Brighton Beach Motor
drome, a few rods away, an aviation meet was going on. Several
times one of the droning machines had gone whirring by over our
heads, so that when the buzzing exhaust of a flier was heard it did
not cause very much comment. Soon, however, the white planes
of "Tom" Sopwith's Wright machine were seen glimmering above
the grandstand. Everyone stood spellbound as he circled the track
several times and then headed out to sea. He was seen to have a
passenger with him. Suddenly, the regular hum of his motor was
broken by severe pops, and the engine ran slower, missing fire badly.
In response, to Sopwith's movements, the big flier tilted and swooped
down to the beach from aloft like an eagle. The terrified crowd
made a rush to get out of the way as the airship came on, but Sop-
with could not land on the beach, but skimmed along close to the
water instead. Suddenly his wing caught the water, and the big
machine somersaulted and sank beneath the waves. The aviators
soon came bobbing up and were taken away in a launch, but the
g 00 accident will not soon be forgotten by those who saw it.
SUPPLEMENTING THE HILLEGAS SCALE
The courage of the panting fugitive was not gone; she was game
to the tip of her high-bred ears; but the fearful pace at which she
had just been going told on her. Her legs trembled, and her heart
beat like a trip-hammer. She slowed her speed perforce, but still
fled industriously up the right bank of the stream. When she had
gone a couple of miles and the dogs were evidently gaining again,
she crossed the broad, deep brook, climbed the steep left bank, and
fled on in the direction of the Mt. Marcy trail. The fording of the
river threw the hounds off for a time; she knew by their uncertain
yelping, up and down the opposite bank, that she had a little respite;
she used it, however, to push on until the baying was faint in her
ears, and then she dropped exhausted upon the ground.
Note. — The first seven of the above compositions, values o to 6.0, were
written during the month of April, 1916, by children in the elementary grades
of the schools in Nassau County, New York. The last three compositions,
values 7.2, 8.0 and 9.0, were selected from compositions which have previously
been published by Professor E. L. Thorndike.^
The value assigned to " The Hunted Deer " (9.0) is that given it in the
Thorndike Extension of the Hillegas Scale.' The value assigned to each of
the other compositions, values o to 8.0, is in each case the median rating of
139 judges who employed as the basis of their ratings the Hillegas Scale for
English Composition by Young People.*
The unit of quality is the median deviation from the median judgment of
the group of 202 judges used by Dr. Hillegas in securing the final values of
the compositions appearing on the Hillegas Scale. In less technical terms,
the unit of quality is such a difference in quaUty as was recognized by exactly
75% of the original judges and not recognized by the other 25%. For general
purposes the bold faced values are accurate enough, although more exact
values are given in small type.*
tempt to eliminate in so far as possible some of these real or
imaginary objections to our educational scales. The supple-
ment to the Hillegas Scale which is here presented does not
eliminate all of the objections, but it does reduce some of them
to such an extent that one may at least hope to find an increas-
ing number of composition teachers making use of objective
measurement.
1 Enghsh Composition — 150 Specimens Arranged for Use in Psychological and
Educational Experiments by Edward L. Thorndike.
" Thorndike Extension of the Hillegas Scale for Measuring the Quality of English
Composition.
* Scale for the Measurement of QuaUty in English Composition by Young Peo-
ple, by Milo B. Hillegas.
* Copies of the above Supplement, in a form convenient for use, may be pur-
chased in any desired quantity. Price, 8 cents per copy by mail; 5 cents in quan-
tities, postage extra. Published by Bureau of Publications, Teachers College,
New York City.
6 TEACHERS COLLEGE BULLETIN
This supplement may be found somewhat less objectionable
than the Hillegas Scale inasmuch as the objections mentioned
above are not so applicable to the supplement as to the scale
itself. It contains no artificial samples. Each sample on the
supplementary scale is as true a representation as can be made
on the printed page of a child's written composition. ^ Each
of the samples appearing in the supplement is sufficiently long
that one may readily obtain a real appreciation of its quality.
Each of the compositions tends to be of the same general narra-
tive type, which fact gives one a simple basis for the recogni-
tion of improvement in quality as one passes upward from qual-
ity o to quality 9. The first seven compositions are upon the
same topic, '' What I Should Like to do Next Saturday," which
is closely related to the type of composition one would use in
writing letters. This is an added advantage, since letter writing
is no doubt the form of composition for which public school
pupils will have the greatest use in their lives outside the class
room. The intervals between the values of the samples in the
supplement are slightly less unequal to each other than are the
intervals found in the original Hillegas Scale itself. The rela-
tive intervals on the two forms of the scale are represented
graphically in Fig. i (page 14).
In preparing the above supplement, the writer made use of
compositions written by the pubUc school pupils in Nassau
County, New York. Each composition in a group of fifty-five
hundred written by these pupils had been rated and assigned
a value^ on the Hillegas Scale by at least two trained judges,
and in case the first two independent ratings differed, a third
judgment had been made by still another judge. From this
mass of roughly evaluated compositions, twenty-eight samples
were selected upon the values of which the first two judges had
agreed. In addition to the requirement that his own judg-
ment and that of the first two judges should agree as to the
quality of the compositions chosen, the writer also made it a
rule not to take more than four compositions rated on the same
sample of the Hillegas Scale. The purpose of this rule was, of
^ The sample valued at 9.00, it should be noted, is the only sample not written
by a school child.
* In these first ratings no effort was made to assign values intermediate to those
appearing on the Hillegas Scale.
SUPPLEMENTING THE HILLEGAS SCALE 7
course, to make sure that there would be in the list two or three
samples of each quality on the scale. In order to meet this
requirement for the upper end of the list, two compositions
pubhshed and partially evaluated by Professor E. L. Thorndike
were included in the list of thirty samples finally selected for
the experiment.
Each of the thirty compositions was carefully mimeographed,
being identified merely by the first letters of the name of the
child who had written the composition. ^ About one hundred
sets of the thirty compositions were prepared, each set being
enclosed in an envelope to facilitate the administration of the
experiment. A sheet of directions was also prepared and mim-
eographed, asking that the compositions be arranged first of
all in the order of their merit and then rated carefully on the
Hillegas Scale. A sample sheet of directions, as it was finally
filled out* and returned to the writer, is shown on page 12.
One hundred and thirty-nine different judges were used in
securing the final ratings given each of the thirty samples. These
judges were all college graduates who had had experience in
teaching and administrative or supervisory school work, and
were at the time of making these judgments members of classes
in educational administration at Teachers College. One group
of 65 judges were members of the advanced or practicum class
studying intensively the problems of educational measurement
during the summer session of 191 6, while the other group of
74 judges were members of a similar group during the winter
session of 191 6-1 7. There is no reason to believe that one of
these groups was by nature any less intelligent or any less able
to distinguish merit in English composition than the other.
The first group of 65 judges were assigned the task of rating
the thirty mimeographed compositions immediately following
four hours of special class work, two hours one day and two
hours the next, in which the derivation of the Hillegas Scale
had been discussed and three or four sets of composition papers
had each been rated two or three times on the Hillegas Scale.
^ Identifying letters were " made up " for the two samples borrowed from
Thorndike.
* The judgments recorded on this sample sheet are copied in the first horizontal
ine of Original Record Sheet No. 2.
TEACHERS COLLEGE BULLETIN
OQ "H
'HOOOO
lOOOOO
OOOlrtO
OOlftOO
0}f ^J
n irtrtioo
Ot^rH^OOl
irtincoNo
t^Tf OWN
OOCOlrtLON
incoco-tN
S on
N COminirt
NOIMOLO
mt^f-iTi>m
tOtOt^MlO
OWOOOIO
}S of
tOooNLn
OincCMN
t^O-^OOOi
o/V
010 05 00 00
!r>iN^ior~
O100N.-IO1
^MMCvJCO
OlOt^-*"-!
Off »7
t^OlTjiNCO
ff.o 97
.-KNCOMCO
lOLnoO'^
(MMOO'-l'-l
(OOli-HOLO
nyj vj
oioomo
TflNLONN
Tf rHiniMOO
0(7 »//
^DC^O-*00
^j "d
OH vyn
OOlONtOOO
to ■'t CO 00 in
to CI C^ 00 t^
00 (CIO SO t^
ocot^oco
COMCMMCM
[^.-(OOrttO
«/Y iO
M3Vi ou
»d "D
»a fH
>M. M3
>PV »/
1,0 OD
^a of
oj o-j
COCOM-"!"*
CO to C^ LOO
CO •*'*C0^
CM CO CO-* 10
00 10000 1>
CM 10 CO 10 in
LOOJtXDlO
M' 10 00 CM CO
t>Ot>t><D
■"tlOCMCOCO
TjlTtOTj-O
lOLO'^lOrJ'
OCM00Ot>
LO •*■*;£! CO
lOt>lO00«3
10 10 CO 10 10
lOOCMtOtO
^CO'-HCM.-l
a)toiot>o
■q.TtcO'^CM
(DOOCMO
-Jf-^CM^CO
OOCJICMCJIIO
LOlOCvlTjiCM
to 00 00 CM CM
to-* --1 10 CM
.-lOtOlOO
COCOCS1C0.-C
Tf to-* 00 00
•* CO CO CO CO
COOCOtDO
■"tCMCMCO"*
OiCMCM-*^
COlO-*-*C0
OCOOOCM
CO-^CO-*-"^"
ocoooo
CO CM CO CO CO
I>CMOLOtD
CO CO •^■* CO
OO-^f-iOlO
Tj-cotrTno
tocMTjiioin
001>10->t.-i
•* CO rococo
0000(0 00
10 CO-* CO-*
t-C0CMlOt>
•*•*-* CO CO
1-itDOOOCM
10 CO CO CO 10
m MD
(DlOlOCM.-(
iom-<ttDco
Ot>OOC0 05
lOlO-*lOi-(
CM (O (J) ■* CM
(0-*CO->t-*
03 H3
-*t>oa)CM
CO •q"(0-*(0
05 --KD 10 00
LOtO^LOlO
OOOOCMOO
■*(D-*1010
t>oc7i--ic:5
in-*-* to 10
^J HD
OKOLOOO
101010(0-*
tOCTllOLOOO
to -^tn 10 10
»a m
(O 000 to 00
CO to to 10 to
t>0(Jli-00
10 tOLO tOLfi
101000CT5
Tttototnto
ns m
OCM-*t>.-l
to (0 00(00
OOC-lOt/50
t> to to 10 1^
tOCMO--ilO
otntoooo
OOtOCMCMO
ooooto
<"S «3
OlOOCOrl'.-i
lOlOOlOO
lOOOOitO OOOlOOtO
t-OOLOtO 1 OOtOOOf-
00 Lo to too
Vtf «7
l>-*t>00)
to to to 00 to
tocM-*(Oto oo-*cri-*-*
inootoo totoioooo
OlCOOtOO
vf pv
O LO <J) o o
0(0(0 l>t>
to 05 CM 000
otooioto
O O O 00 o
CMOOOCn
otooooto
IS »il
oooo-*io
00000
OIOIOOO I tooo-*
OOOOOtOO OOOOtOOiOO
«S »H
tD-*Orf<-*
000 0)000
-"tOOi-ltOCM UOOOCM'-i
OitOOOOO OChOOOO)
SUPPLEMENTING THE HILLEGAS SCALE
OOCOOtO —oooo
Ortt-Ttin
tOLnoNTf
CMi-iinooT}'
LnMoo— if-
PONOOt^f^
OOOOIOCO
MCOtNlO
MOiwoo ■<toot^ino)
mmtooiw
IM (M -H
COMTflOOO
^tOOOOlN
CT*'«f «0
MtOO)ON
omotof~
XI ^ lo r~ in
OOO-^LO
inoootn
0)03 05 CO ID
INi-iMM^
OOt^OOOO
00 -H LO LO (N
-^NCO.-iCO
I>LO.-iOOOO
NCOMi-irH
M-*iMrqr-(
^ CO CM
OOCOIOLOC^
riCMCOCO-*
«3t>C>J!£)C0
i-'i-iCOM.-i
C^OOOOIO
rjcgM— ICO
■* 00 ^CtTO)
NCMM-*CO
MCOCOIOM'
•-(COCOMCO
oo^]O)coc^0
COC^C^-HCO
CJ —■ "3- PJ CO
LOOlOOOOCM
CO CO CO CO ;d
CO-^PJCMN
OllOCOOO^D
—iMCO-hCv)
NOOCVJOlCO
in --I CO .-1 ■*
ocooKOLn
CO CO CM CO-*
(Ot~-t^C0t>
CM CM CO CO CO
0)0;OCO^O
CM'^LOCO'^
■•s"too<rio
t^-* 00 coo
CO'^ ■^LOtO
COOilO'^O
O.-1-.^COLO
•^•^C^CMCO
10 CO CO -sf'-fl
Ot^LOrHOl
coTfcoioin
incO^CMlD
CMt^OlOOCJ)
CO CO CO CO CO
CM CO CM CO Tr
cMLnoco-*
COOOOtlO
COCOUJCOrj-
t^O-*^DO
CO-^UI-^LO
Ol-HCDCMt^
<D0OtD!£)i-l
cOTfcg-ttco
incMcoco-^
■"f O CM CO 'J'
O)-^00-*CM
CO-* CO com
tOC0U5LO;D
Ol-*C0tD-*
COCO-flOLO
LO-t ->J"*t^
Ln-*oo>(N
CM 1/5 000-^
CO-* so COLO
OlCOCMt^OO
t^Ot-t^LO
-* in •* '^ lo
lOtCUDSOCM
CO CO CM CO-*
SOOCOt^^
coTfinin ■*
(CCM05CMO
in Tp^ln to
in in so-* to
t^t^t^inoo
CM CO-* CO in
oooocjiin
■* in to CO to
t^ CO in in in
cMin-HOoo
-* CO to in CO
in in to to c^
-* ■*tointo
CO) to 0)^00
-* •* msoto
OOCMOitOC^
C0-*tO'*tO
CM to CM t^ in
so-* in in in
t^CMOiCJlCM
co-*Lnr)'in
CM in t^ 00 00
to into in in
in •*t^ 00 CO
CO to in in t^
05t~-int^cM
in -* in in to
CO r~-* to CO
Tl<tOtO-*t~
— lOCMCOi-i
-*inc^tDLn
oot-~t^o ■*
in-*totoc^
to f~ in 00—1
t- to in in t>-
00 CO CO 0)00
■v-^tomin
CO O) --I c-J to
to CO c^ to to
tot^ooinin
to to to to r~
C00-Ht^05
t^OOOt^CM
soinin WC--
oot^toooo
oootot^oo
intot^i>to
tntot-^CMt-*
t^ t^ t> c^ t-^
C^CSt^CMt^
c^int>-c^c-»
to to to 00 00
t^OlOOOCM
-*-*tD00 0O
ot^ t^inoo
CMt^inoc^
00-*tO t^r-
CMt^CMCMOi
t^tot^ooo
t^OOOOTl-O
•*0)0ino
'5'-*CJ)t^00
OOt>tDt>00
tOr^OOOt^
-*S0tOOltO
CMCMinO'-i
CO 00 to co^
00 1> in 00 CO
c^t^t^tooo
oincM— I— I
t^I>t^O)0)
to CO CM CM 00
tooooot^t>
oocor^-HCO
to O) to 00 00
t> in to 0)00
OOtOt-'CJiCO
in in t^ 00 c:^
inocooco
t^O)00t^00
McocMinin
l^ 00 O) ■* 00
OCOCOOO)
oioomoor^
t^ CO CM o o
Sto «
""oOjs
1) I,
x: 1
m aj 3
o'^ c
in cj w
c:.2 ^
^ Mt«
"2i:
<U w O
> N
HJ
V £: j:
tu M^
SS ™
.9mW
S^ '^
ECi ij
o aj a
" c '"
" £ t^
=" DO
c!3 ij
O C3 cu
rag"
"«o)-
•«"Ln
o ♦J
c
•^to
t«
3 3
CO -0=0
c:
Bl
Sm
o
k5
J3
Ma
-
b-o
1
js »
r/l
t.t>-
C
u
CQtO
o
o
h «
m
^
<U Q
0
ri^
u
o>i
b
0.x
u
lO
TEACHERS COLLEGE BULLETIN
OQ ^U
Oi-iOTrLO
OOOi-iTtO OOOOIO OONC>3-H OOIOOO
o;/ ^j
i-HinLoooiM
OOl-^OO O-^NNOi CSlOOt-r^Tl" tDON'-i<£>
H o}{
<woomt^
i-nocOfHin
lOlrt— irf<r 00L0.-1OO r-UDlrtcOCJ
}S of
OMlrtMin
OOOmt-HM 0IN«5-<^0 ^tOONlO
D/V
ooinoo
(N M .-H
o^ooin
LnoLOi-HO
^^DCOCOO 00t^Ot>M OlONOHn
Off »7
rtrH coined
O)0in.-n£>
ff,0 37 SS
«/v »</
"(7 "H
^A "d
CO to 00 00 03
r-iOOCMtO
Oi/^ OJ^
■-H Tf ,-1 (M CO
.'/V ■'9
N^IV OS
oooojt^rj
LOIOC^NCO
»<:/ OJ
COCOi-iCJ>M<
Ttin-*i-ico
»ff .»iir
OOCSlt^MOO
t^-* (DID 00
CO CO ■^rt CO
«/tI V3
•*-*coLnTj<
ooo<DO<r>
«/v »/
Tj'tDCOOOO
•q- in tt tc T^
Oi-it^tDN
inrf rf COM-
CD -ooint^
■^tDTfCOrC
I.O 03
(D05Tt"a'ir5
Tl'Tj'COtO-*
^a of
COt^tDincO
1/5 LOCO to-*
.-1,-HOOOOO
imntocoin
--ic<ii>ooco
•* Tt LC t> Tj-
oo-*oiooai
lOlD'^LT)-*
lOi-iintrti-i
■"tCOlOCOTj"
oj o-j
■1(7 V3
OILOOOOC^
■^LOLOLnin
lOOOCONtD
■*IOCOOOM<
t^t^OOl^D
co^tonn-
fj O) .-I t^ (J»
(DinTj-tDCO
COOtXT)--!
1/5'^ ■^^D-^
.-HOOOONN
O'S'Lnt^io
NOOCOO-H
LO t^ -^ tn LO
^LOLOCOtD
N(D-HMCO
Tj-cgLOTj-LO
oONoomco
lo ■* LO T)< -a-
oj HD
LO<DLOI>'t
LOlMOlLOr-l
LfltDCVUOCO
.-uMai-HCV)
CD t^ 'J' CO to
ioooo^oot~
LO COLO -a- CO
■(^ HO
OOOOOOOl
^(DtOOO'*
lOLOt^l>00
LOLOtDtDtO
tXDLOtDCO
^OOOCOO
001>-*CO!D
»ff VJ^
■q-LOLOOLO
00 <M to COLD
(DLOtDt^LO
t>t>LOCTit>
to to to to to
"s .«;¥
03 t^ O) "tf tj)
oirjc^otD
a^r^Ln-^io
t>l>t>00LO
OCOOOtDCO
t^OOOOt^t^
t^t^tOLO'^
ffig n;5
t> t^r-oico
tOLOTj'ON
t^t^oooot^
t^i>t>ooc^
t>lO.-HLOtO
ooooooot^
t^NMLOO
t^OOtDl>00
OOOOLO t-
l>00LOtOt>
OOLOtDO
ootoc^r-tD
HS "7
OOMC^CO^
t>00<DOO0O
t3>OOOLOCO
C^tvC^OOO
COCOI>.-iO
OOOC^OIOO
OOt^OOLO
OOC^tOt^LO
vf pv
OCOO-*LO
O0 00t-O5t^
Ot--Ttt^00
cooot>oot^
00 00LOtDL>
cgLONt^i-i
00LD00t>tD
>S »H
COOOOOLO
0Ot>O0tJ)tO
OOOOLOCO
OOOOt^t^tJl
Tf OONCOIO
oot>ooor~
^coot>ai
00 00 00 00 1>
lOtOO^LO
0> 00 05 00 00
OrjtDCqtD
OlOl^tOtD
-JOCO'-ILO
oot^oit^to
*S »H
tOOCOf-N
t^oooooc^
t^LOLOOOl
t- 00 00 05 to
f-H'S'LOLOLO
t^i>O>00i:^
SUPPLEMENTING THE HILLEGAS SCALE
11
2
o
M
O
O
U
o
H
Q
W
en
<
CO
<
i-l
w
0(7 «y
MWOO-*
OOIOCON
i-H 1 I
-^OOCDO
OOOON
o-*ooo
^oo-o
OOOO
OJV -fj/
NNioirsLn
iOOTi't^Ln
COIONNCO
NlOrHt>00
OtOLOt^OO
10OCML0 05
coco rH
OOtO^M
y 0^/
■HMCMOOO
rH 1 rHCO
COrtOlMOO
CTiMrHW*
■^OOOrHLO
C^JLOOOt^OS
rtCO
^Nt-tNt^
Ot-O^
fS of
O WMO5 00
lO-*COOO
COtOrtOOO
t>OtOCMLO
•-ICSI <N
tOlOOt-O
^<N
lOO5 00rt
opV
000-*OLO
OO'l'C^lOOC^
OOOlOCOLO
^cqOMOi
totot>ot>
'* 0)05'* 00
to 00 to to
og »7
(Mr-i com
tnoiLOf<iO
cO'Hcocorg
TftOOOOCM
OOtOOONlO
rHcsi^pgcM
COOCO-^M
c<IOooo^^
CMCMrHrHC<l
lO CM to CO
CM-HCM
ff.0»7
ooocsiinoo
UDLOLntDM
COCg (NW
00t0Ot>L0
■-ICOCJrH.-l
LOOOCOOIN
^COPJOOO
f^cor-irgco
OtOLOtO*
OtOO)00
rHrtCM
«/V "c/
un'-iffiMoo
CO CO CO CO cj
001010 05 00
OI>lO-HO
CMNlNcgcO
to 00 to CO 00
M ^C^4rt
O00tOO5 1>
corqrtr^cM
OCMOOt-
COCMCM
Off 0//
lOOOOOtO^
rHCJCOCOIN
tocot^ot^
IM-*CSIC^1^
t^ 00 LOCO 00
rtcortc^csi
CM-^C^LOO)
lOCO-HMrt
cMtoor-io
*CMrococM
ooot-
r-H'*CM
'^«5
i-HOlMl>00
OOOOOON
rqrtcv)ooc<i
to to o o to
cQcoc^c^cg
■*005<010
iNCOrtNrH
Tf tOlNt>tO
'*'*'* CM CM
cMtortin
CO CM CO rH
on ofv
i>ooot^m
cocj.-iTj'co
0)LOCJ)f-CM
CO^MCMCSI
t^ioooooo
NcO'-i t^co
tOOOl^O)
^c^c^co
lO00'*t>LO
CM CO -H (M fj
Olo—iloco
CM'*'* CM CM
coototo
■*rHCO.-l
1
>fV '■9
MMNCOtJ'
M'-cOlOOt^
rfTtcOCOM
■*t>LnrqtD
l>LO00tOt^
CO'^CONCM
00 to 00 too
rHCO-HCO!N
lO00-*00'*
CO CT* -H CO
tOCOf-t-
WCOCO-H
N^iv on
ot>tnoCTi
tfCMTj-OCO
COO'* eg-*
t^COtOC^'*
CO 00 CO-* CO
^totooto
'*t^rH0O^
f-t-ooco-*
•*-**-*'*
tooooo
coc^]'*^
CO
»d oj
OOOO'HOO
CO -g- CO (C -q-
OOCJlOOO
coLOTpLnco
LOooO'^Ln
MLOTf IN
—1'* 05 00 05
pqcocoioco
Cg^t>rtCO
-*t>COC0N
005COtOCM
■*-*'* CO*
LO to 00 to
CO CM CO CM
Q
»;
o
o
w
»g }d
OCT! 00 pq CO
'J'COlOt^CO
OHD^tOOO
M-*CO(N-*
C^tOtDOO
Tt^'^lO'*
0)M05C0t-
rHOOMCOCO
t-O'^LOtO
to to LOCO CO
LO(N'*tD
■* CO LOCO
>M MD
CO CO LOOM
Tj-cMcoLOun
lO-*-H00-*
lO-^Tf^DCO
Lnt^ot>io
CO'* CO CO CO
OOOCOt>LO
C0t>-*IMCO
CMOMOO
NtO'**Ln
lOO)CMNL0
•*lOlO-*CO
C0lOO5'H
CO*'* CM
}Vi »[
O (C Ol 'sD Tf
incT>^Lnt>
Tj'^LO'3'CO
t^LOLnoico
core CO COIN
05 00'* 00 to
rt-^TCc^co
ot^rtoor-
CMtO^CO-*
tot-t-ioco
TriO'*'*co
CMOtOO
■*LO'*CM
o
o
1.0 oj
00t>P3CT>O
Tj- Tj- -J. !£) in
■^OOCOrfO
Oioorjooo
■*tO'*C0'*
t-.t^^lOCv)
■*LO'*'*'*
to 03 to t^ t-
lOLO-*C0tO
OOrtOOtOlO
Lot- to to-*
t-toooio
-*to*co
^a of
lOcoLniOLfi
t>-*ooai«o
coco-* too
CO t^-* COLO
C0Ot>00C0
IOlOtI"*'*
to CO t^ CM to
LO 00'* CO to
000 0501--
t>to-*to^
OOt-O)
■*C0'*^
03 01
in Tj. (£) to U3
I>-CqOC0C^3
O>CT)00r^LO
COIOCO-*'*
OPJIOLOOO
tolr^-^iNCO
CMOC0t'~00
tO00**lO
LOCOOOt-CM
to to to-* CO
■*tooto
CO to to CM
^a ID
.-irjtoooin
Tj< Ti- ■* tD ;d
CO^OOCOCTl
■*Ln untried
in ',o CO o lO
^to-^coto
MOOCOf-
t>tO'*'*C0
'*lO'*'*LO
tooo*coio
CM CM CM 00 to
LOtOtOCT*
'* CO to CO
OD M3
coLncj)t>.^
OOtOCTlOlCO
M'tO'3't^-*
'^intDrg'*
lOt>^'*'*
totooooto
C^t^-^tOrP
tO'S'lOCMCM
to 00-*'* to
CML0C0O5tO
to to t- lO to
t-ot-t-
tO to to CO
^J HD
tOOOCOOOl
tOLn<£>tr~(D
■^CMtflt^CO
LOIDLOOCO
00'>*t^CO'*
coc^-^-^c^a
O-lOLOtOt
COIOOOOO
i>t>tointo
OtOr-^OOLO
tOtOC-LOtO
OOtOLO
t-t-lO'*
»a m
tOLOtOtXD
OOOOOt^
<0 [> t£i tr^ rji
OCOLOOCO
■^•0^*00
COlOlOtO^
tXOLOtO ■*
t'^ to 00 05 00
to 00-* CO c^
tO'*Ot>0
to to to tot—
OLOt-O
to to to '*
ns ?fV
Ol C-- O 'X> o
COOOOO-H
f~t^t»00t^
oi>o-<to
t^t^LO-^c;!
^OOOCOf-
t>05LOtO*
00O5t-l00W
00C^tOLOf~
LOOI>t-00
t-oot-c-to
t-LOt-lO
t- t-t- to
■WS "3
LOt>-mLnco
lO(£lt>t^tD
loootooiin
OOOOOILO
y5t>tDLOI>
lOCOLOLOlO
00 001>t>t>
O^L01>rt
00 05 LO to 00
t-005t-tO
t-t-totot-
00<J50
O)05tOO)
MH «7
ooo t~t^-*
CD00t>t><£)
O O) 00 C^J 00
LOcorjoo-*
(OOOCOIOOO
C0 05t>l>.05
00t>0t01>
t0 0001>05
t- 00 tor- 00
CM CO CM to CM
00 00 00tDt35
00 LO to CO
OOOOt-O)
of pv
t> 00 00 1.^ «D
00OOCJ5 10
'*CT>L0t^O
t>t><oiot>
■^ooor-oo
00O001>t>
'*0050CM
05 05O00 00
C^LOCDOt-
OOt-OOt-t-
CO CO 00 't
05 O) t— 05
IS »n
T)<O00 00I>
NOLOOLO
oooooocjit^
I>0001>0>
tOC0N'*l>
00 05 00 to 1>
t-'*MCOCM
C0O5CO00O5
OOCMCOO
00 0)0)00 05
to CO (35 to
00 00 t- 00
..s*//jjgSSSf^
oOLnoocooi
0-5t"LOOCO
00 05 Ul tool
OLOOOtOO
oo05C^Lnoo
tOCMLOOCO
00 0)0) £35 00
CMCMt-C0 05
t-t- 05 05 00
COLO 00 to
00O)tOO5
12
TEACHERS COLLEGE BULLETIN
In connection with this rating of compositions in class, certain
compositions had been read aloud to the class and a general
discussion had been held, dealing not only with the merit of
Directions
Do all of this work independently. Do not let your judgment be influenced
by anyone or by anything other than the quality of the compositions themselves.
First:
Arrange the thirty compositions in the order of their merit, giving the best
composition rank number i, the next best rank number 2, and so on down to the
poorest, which should receive rank number 30. Do not give any two composi-
tions the same rank.
Record your judgment in the proper place on the blank below by writing neatly
(or printing, if your writing is quality 9 or less) the identification letters of the
composition opposite the rank number to which you have assigned it.
Second:
Determine as accurately as possible the position of each of the thirty compo-
sitions on the Hillegas Scale or on the Thorndike Extension of the Hillegas Scale.
You may assign values intermediate to those given on the scale if you feel rea-
sonably sure of your judgment. (You may, for example, wish to assign a value
of 05 to one composition, 10 to another, 15 to another, etc., even though there
are no samples on the scale between o and 18.)
Record below the value you assign to each composition, and return your record
and material to Mr. Trabue before consulting with anyone regarding your judg-
ments.
Rank
Number
Composition
Lettered
Quality on
Hillegas Sc.
Rank
Number
Composition
Lettered
Quality on
Hillegas Sc.
I . . . .
Re St. .
82
16
Ri Be
^8
a. . . .
. ...AdJa....
80
17
... Ca Pe
. ??
I. . . .
. . . .He Si . ■ ■
7n
18. .. .
4. . . .
.... Eu Sw . . .
78
10. . . .
Ro McN.
2C
?. . . .
... .LuRh...
7c
20
Ha Da
2d
6
....MiSu....
60
21 . .
... Qu Fr
2 ?
7. . . .
. .. .Jo Dr....
c ?
22. . . .
.... Ma Ro .
21
8... .
. .. .ChCo....
CO
2^. . . .
Ma
20
0. . . .
.... Lo Co ....
40
2A
Jo Si
. 10
10. . . .
....ChFr....
48
2K. . . .
Ro R .
18
II. . . .
... .ThBe....
47
26
. . . .Pa Mu...
17
12. . . .
CoO'T...
46
27
. . .Le Bo
. . 16
i^. . . .
....ChDr....
4c
28
Le O'B .
Id
14. . . .
. ...JeMi....
44
20. . . .
....FrKo....
17
IK. . . .
Ch Wi. . . .
42
^0. . . .
. . .Ru Do . .
0
SUPPLEMENTING THE HILLEGAS SCALE
13
the particular compositions read but also with what constitutes
general merit in English compositions. This group of 65 judges
will therefore be referred to during the remainder of this dis-
cussion as the " trained group." The ratings given by these
judges on each of the thirty samples of composition will be
found in Record Sheet No. r.
The second group of 74 judges were assigned the task of rat-
ing the thirty mimeographed compositions before any discus-
sion had been held in class either of the derivation or the use
of the Hillegas Scale. This group of judges were asked to indi-
cate on their record sheets whether or not they had ever used
the Hillegas Scale before, but some of them neglected to make
any statement regarding the matter. Where no statement
was made, the writer grouped the records of those who had
been in attendance at Teachers College previous to this year
with the group who stated that they were acquainted with the
scale, while records made by new students were grouped with
those who reported that they were unfamihar with the scale.
The judgments of the group who were " unfamiliar " with the
scale are shown in Record Sheet No. 2, while the judgments
of the " familiar " group are shown in Record Sheet No. 3.
The median 1 rating of each of the thirty compositions is
probably the most accurate measure of its real position on the
Hillegas Scale.
A graphic representation of the median value of each compo-
sition is shown in Fig. i. An ideal scale would, of course, have
one sample at o, another sample at i.o unit above o, another
sample at 2.0 units above o, and so on. Not enough of the
samples used in this study have values at exactly the right points
to make possible such an ideal scale, but as close an approxima-
tion to the ideal as was possible has been made. The values
of the resulting supplementary scale are shown in Fig. i, to-
gether with the values of the original Hillegas Scale and the
values of all the compositions from which the samples of the
supplement were selected.
'The median rating is the rating above and below which there were an equal
number of ratings, — in this case, 69 judges rated higher and 69 rated lower than
the median rating on each composition.
14 TEACHERS COLLEGE BULLETIN
The reader will observe in Fig. i that two of the thirty samples
are very near quality i, two are very near quality 5, and two
are very near quality 8. In so far as the median ratings are
concerned, it would apparently matter very little which of the
two compositions in each of these three pairs was retained as a
part of the supplementary scale. In order to determine which
one should be used in each case, the distribution of ratings on
each composition of the three pairs was more carefully studied.
It was found that although both compositions He Si and Re St
have a median rating of 8.0, the judges had greatly overesti-
mated or greatly underestimated the value of Re St in fewer
HILLEGAS p
30
Samples
10
Fig. I. Values on the Hillegas Scale of thirty compositions from which nine
were selected for the Nassau County Supplement
cases than they had the value of He Si. In other words, the
judges had agreed as to the value of composition Re St much
more closely than they had agreed as to the value of He Si.
In a similar manner, although both compositions Ch Dr and
Lo Co have a median rating of approximately 5.0, the judges
agreed more closely as to the value of Lo Co. Compositions
Ma and Jo Si, likewise, are both very near quality i.o, but
there were somewhat fewer extremely bad judgments on compo-
sition Ma. It seems quite evident that to be of the greatest
usefulness a scale should be composed of compositions about
whose quahty there is the least possible disagreement. In the
three cases, therefore, where two compositions fall at approxi-
mately the proper point on the quality scale, that sample was
used which showed the smallest number of very bad judgments.
SUPPLEMENTING THE HILLEGAS SCALE
IS
In order to see more clearly the significance of the above
paragraph and to understand the next few pages of this dis-
cussion, the reader should recall how the values of the Hillegas
Scale were established. One unit on the Hillegas Scale is just
such a difference between composition x and composition y
as would cause 75 per cent of Dr. Hillegas' 202 original judges
to declare that x was better than y, while the remaining 25
per cent of the judges declared that y was better than x. If
50 per cent had said that x were better than y and the other
50 per cent that y were better than x, then we should have
been safe in saying that x and y were equal. A difference
noticed by 60 per cent but not noticed by 40 per cent would
probably be a real difference, but it would not be a very useful
difference, for 4 persons out of 10 would not be able to recognize
and employ the difference. A difference which is noticed by
3 out of 4 persons is, however, a really useful difference, and it
also agrees with some of our statistical concepts.
Dr. Hillegas assumed, and there is every reason to believe
that the assumption is perfectly sound, that judgments as to
the quality of a composition will be distributed symmetrically
on either side of its true value, — that just as many will over-
estimate as under-estimate its value, and that large errors in
either direction will balance each other and be much less fre-
quent than small errors. If the true value of composition x
is represented in Fig. 2 by the point marked o on the base line,
the judgments as to its value would, according to the assump-
tion, be distributed both to the right and to the left of o in
diminishing frequencies represented by the surface enclosed
between the heavy curved line and the base. This symmetrical
surface representing the judgments on composition x, whose
true value on the base line is o, is the well-known friend and
servant of the statistician, known as the Normal Surface of
Frequency.
Another assumption made by Dr. Hillegas is that the variabil-
ity of judgments on one composition is exactly equal to the
variability of judgments on another composition. In Fig. 2
let us suppose that a second composition y has a true value on
the base line which is just sufficiently far below (to the left of)
the true value of composition x that exactly 75 per cent of the
i6
TEACHERS COLLEGE BULLETIN
judges in distribution x assigned to x values above (to the right
of) the true location of y. Then, if judges vary just as widely
on one composition as on another, we should expect to find 75
per cent of the judges in distribution surface y rating y as below
the true value of composition x on the base line. The sta-
tistical name for the difference between the true value or median
(50 per cent) point of a distribution and the point on the base
which has 25 per cent of the distribution on one side and 75
per cent on the other is the Median Deviation (M.D. or P.E.).
The second assumption may therefore be restated to read,
" The Median Deviation of a distribution of judgments on one
composition is equal to the M.D. of a distribution of judgments
on another composition."
-5-4-3-2-10123456
Fig. 2. Assumed equal variability of judgments on all compositions
The statement may now be made, which might have been
made at the very beginning of the discussion for those who are
familiar with statistical terms, that the Median Deviation is
the unit of difference employed in the Hillegas Scale. The
sample composition which on the scale has a value of 2.6 was
judged as better than 2.6 by 50 per cent and as poorer than 2.6
by the other 50 per cent of Hillegas' 202 judges. Twenty-five
per cent of these judges would have judged it as poorer than a
sample whose actual value was 1.6, and on the other extreme
25 per cent of these judges would have judged it better than a
sample whose real value was 3.6. We may assume that if
Dr. Hillegas' judges had included in their Hst the composition
Re St used in this study, 50 per cent of them would have rated
it as better than 8.0 and the other 50 per cent would have rated
it as poorer than 8.0, for the median of 139 judgments places
this sample Re St at 8.0. If the M.D. of judgments on one
SUPPLEMENTING THE HILLEGAS SCALE
17
composition is equal to the M.D. of judgments on another, we
may further assume that 25 per cent of Dr. Hi]Ie;:^as' Judges
would have rated Re St as poorer than 7.0 and another 25 per
cent of them would have rated it as better than 9.0. It is
quite surprismg, therefore, to find that only about 10 per cent
of the 139 judges used in this study rated Re St as low as 7.0
and only about 10 per cent rated it as high as 9.0.
If the base line of the two curves in Fig. 3 is allowed to repre-
sent M.D. units on the Hillegas Scale, the low rather fiat curve
encloses the surface representing the general distribution of
judgments to be expected (according to Dr. Hillegas) for any
Re St
Gsneral dls-
trfbution &
scale
r Re St ao&le
— HeSl ecale
Fig. 3. Small variability of judgments on one composition
sample of composition, while the narrow high curve encloses
the surface representing the actual distribution of judgments
on composition Re St. whose median value was found to be 8.0.
If the M.D. of the Re St distribution were used as the unit of a
scale, in place of the M.D. of an assumed distribution common
to all compositions, note how much shorter these scale units
would be and how many more units would therefore be included
between any two different points on the scale. Fig. 3 also shows
the M.D. units that would result from using as the basis for a
scale the distribution of judgments on composition He Si, whose
i8
TEACHERS COLLEGE BULLETIN
median value was also found to be approximately 8.0. If the dis-
tribution on He Si had been shown it would have been only slightly
higher and less wide than the general distribution assumed by
Hillegas. Such large differences in the form of surfaces of
distribution of judgments on two equally good compositions
were certainly unexpected. They tend to make one critical of
the assumption that the variability of judgments on one com-
position is equal to the variability of judgments on any other.
A rough distribution is given below in Table I of the 139
actual deviations from the true value for each composition in
the three pairs of equal samples mentioned above.
TABLE I
Composi-
tion
called
Median
value
DiSTRIB.
FROM
OF Amounts of Deviation
Final Median Value
Total
0 -S
to to
•4 -9
i.o i.s 2.0 2.5 3.0 3.S
to to to to to to
1.4 1.9 2.4 2.9 3.4 3.9
devia-
tion
He Si
Re St
8.01
8.00
47 30
62 30
31 IS 9 4 2 I
37 S 2 I I I
139
139
.82s
.487
Ma
Jo Si
1.06
1.03
S3 67
SI 60
18 I
22 s I
139
139
.627
.S68
ChDr
Lo Co
4.98
4-97
28 43
42 46
30 23 12 21
27 IS 6 2 I
139
139
.900
.766
' The median of these deviations was calculated from a much finer grouping than that
shown in this table, hence the unusual M. D on Ma.
In connection with this table it was observed that the M.D.
of judgments for each of the six compositions was less than one
unit (i M.D. of Hillegas' judges). This fact had in it a sugges-
tion that possibly the 139 judges used in this experiment agreed
with each other more closely than the judges used by Dr. Hille-
gas agreed among themselves. To test this possibility, the
method used by Hillegas on the judgments of 202 judges was
employed upon the judgments of the 139 judges used in this
study. The resulting calculation is shown in Table II.
Table II, beginning at the top and reading across horizon-
tally to the right {rf. Original Record Sheets 1,2, and 3) reads
as follows: Comparing the values given by each individual
SUPPLEMENTING THE HILLEGAS SCALE
19
TABLE II
Result of Applying Hillegas' Method to 139 Ratings on the
HiLLEGAS Scale
No. of Records
Amount
showing
Percent-
"Poorer"
Amount better
Relation of
age
( ^ )
than Ru Do
"Is "Is
"Better"
\ M.D. /
(i=M.D.)
Poorer" Better"
Ru Do to Ma
^33 6
4-3
2-55
Ma is 2.5s
Ma to Le Bo
131 8
5-8
2.34
Le Bo is 4 . 89
Le Bo to Ma Ro
120 19
13-7
1.62
Ma Ro is 6.51
Ma Ro to Ca Pe
117 22
15-8
1.48
Ca Pe is 7.99
Ca Pe to La Co
"4 25
18.0
1.36
Lo Co is 9.3s
Lo Co to Th Be
119 20
14.4
1.58
Th Be is 10.93
ThBe to Mi Su
129 10
7.2
2.17
Mi Su is 13 . 10
MiSu to Re St
123 16
"•5
1.78
Re St is 14.88
judge to compositions Ru Do and Ma, 133 gave Ru Do a
lower rating than Ma; while 6 of the 139, or 4.3 per cent,
gave Ru Do a higher rating than Ma; converting 4.3 per
cent into terms of M.D. by means of the table used by Pro-
fessor Hillegas, there are 2.55 M.D. units between Ru Do
and Ma. Similarly, 8 of the 139 judges, or 5.8 per cent, decide
that Ma is better than Le Bo, so it is therefore 2.34 M.D. worse
than Le Bo. Le Bo being 2.34 units better than Ma, which is
2.55 units better than Ru Do, we must conclude that Le Bo is
4.89 units (2.55 + 2.34=4.89) better than Ru Do. It will be
observed that Re St, the best composition, is 14.88 units above
Ru Do. It will be remembered that the median values of these
two compositions on the Hillegas Scale differ by exactly 8.00
units. The difference in quality between Ru Do and Re St
actually remains the same, but if one accepts the M.D. of the
139 judges as a basis he finds 6.88 units more between them
than he finds if he accepts the M.D. of Hillegas' 202 judges.
This difference is shown graphically in Fig. 4.
RD M LB MR CP
• • • • •
I I I I I
LC TB
MS RS
I I I
Median RATINGS vs.
Calculated VALUES
RD
LB
MR
CP
LC
TB
MS
RS
I I I I I I
O 1
Fig. 4,
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Linear projections of median ratings and of M.D. values for samples
included in Nassau County Supplement
20 TEACHERS COLLEGE BULLETIN
The discrepancy between the two groups of judges may be
due to the fact that the 139 judges had the opportunity of using
an objective scale (the Hillegas Scale) in their arrangements
of the thirty compositions, while the origina 202 judges had no
common objective measure of merit in composition. The 139
judges may agree more closely with each other because they
were all " school people." The fact that the thirty composi-
tions were all of the same general type may have been partially
responsible for the greater amount of agreement as to differ-
ences in merit. Probably all of these factors combine with
less obvious causes to produce greater agreement among the
139 judges in this experiment than existed among Hillegas'
202 judges.
In order to see whether the four hours' discussion and use of
the Hillegas Scale in the summer session class had resulted in
any increased ability to agree as to composition quality, Dr.
Hillegas' method was used on the 65 " trained " judges' judg-
ments and on the 74 " untrained " judges' judgments separately.
The results of these calculations are shown in Table III. The
reader will recall again that Re St is only 8.00 units above
TABLE III
Measured by
A. Median on Hille-
gas Scale
B. M.D. of 74 Un-
trained Judges
C. M.D. of 65 Trained
Judges
Distance above zero of samples in Nassau County
Supplement
0
0
^
C)
0
CQ
(<
0.
c">
cq
^
c^
0=;
0
c3
0
-1
S
%
o 1.06 1.92 2.80 3.84 4.96 6.00 7.21 8.00
O 2.22 4.17 5.41 6.87 8.33 9.71 11.66 13.39
o 3.2s 6.50 8.79 10.30 II. 55 13.38 15.88 17.71
Ru Do, if we use the M.D. of Hillegas' judges as the unit, while
it is 13.4 units above Ru Do if the " untrained " group are used
as the basis, and if the 65 " trained " judges are used as a basis
the difference is increased to 17.7 units. As was said before,
the " trained " group were probably not more capable by nature
than the " untrained." Four hours' discussion and use of the
scale had merely made them agree much more closely in their
SUPPLEMENTING THE HILLEGAS SCALE 2 1
judgments of composition quality, even though they rated the
thirty compositions without ever having seen any of them be-
fore and without conferring at all about them. These differ-
ences are shown graphically in Fig. 5 for each interval between
samples of the supplement.
RD-M M-LB LB-MR MR-CP CP-LC LC-TB TB-MS MS-RS
1 ^^ 1.0c Hi M ^m -88 ■^1.04 ^■■1.12 ^HililA ■^■1.21 m -79
Fig. 5. Influence of training upon discrimination of differences between samples.
In concluding this discussion of the derivation of the Nassau
County Supplement to the Hillegas Scale, the writer must
explain that he does not advocate discarding the Hillegas Scale.
It was impossible, before the Hillegas Scale was devised, to know
that judges would be more variable on one composition than
on another. Now that we can measure such differences in
variability fairly well, we should get together a large number
of compositions upon which judges do agree very closely. Just
because the median judgment places a sample at the right point
on the quality scale is not sufficient excuse for using it. Only
those samples upon which people agree closely should be used.
In order to get a very perfect scale, it would possibly be well
to train the judges until additional training would not result
in greater agreement among them. If the variability of human
judgment is not a constant on which our scales for comparison
may be built, possibly the variability of the judgments of col-
lege graduates having ten hours' training in doing this work
would more nearly approximate a constant for reference. For
all practical purposes at present, however, some form of the
Hillegas Scale, crude though it is, will supply all our needs in
this direction.
The need for the second supplement, a tentative statement
of standards to be achieved in English composition, was empha-
sized very strongly by the differences between the results being
obtained by the same grades in different schools. To find fourth
and fifth grade classes acliieving results distinctly superior to
eighth grade classes in a neighboring school is rather surprising
22
TEACHERS COLLEGE BULLETIN
until one stops to consider that up to the present no one has
told these teachers what quality of English composition to
accept in their classes. Teachers probably determine their
standards most commonly from the compositions pupils hand
in during the first few meetings of the class. In very rare in-
stances samples are preserved from year to year and used as a
basis for grading compositions from later classes. In the large
majority of cases, however, a new teacher has no way of learn-
ing what to expect or require. The random results which she
obtains are illustrated by the records of the eight classes in-
cluded in Table IV. Two of these classes are fourth grades,
two are fifth, two are eighth grades, two are senior classes in
high school, and all of them are in the same county within ten
miles of each other.
TABLE IV
Records of Eight Classes
No. of compositions rated on sample of
Median
75%
In
value
Total
class
No
o 1.8 2.6 3.7 4.7 5.8 6.7 7-7
Quality
A
3 27 9
39
3.80
3-41
B
I 14 8 I
24
2.94
2
55
C
6 8 12 4
30
6.37
5
48
D
I 8 19 4
32
5-66
5
16
E
S 23 19 I
48
1.99
I
31
F
34 7 I
42
3-8i
3
48
G
6 6 I
13
6.38
S
84
H
12 17 5
34
3-46
2.87
A and H are fourth grades, E and F are fifth grades, B and C are eighth grades, and
D and G are fourth year high school classes.
The above table reads from left to right as follows: Three
pupils in Class A wrote compositions which were scored on 2.6,
twenty-seven pupils wrote compositions which were scored on
3.7, and nine pupils wrote compositions of quahty 4.7, making
altogether thirty-nine compositions, half of which were of
quality 3.8, or better, and 75 per cent of which were of quality
3.41 or better.
Notice in the above table the difference between the two
eighth grade classes or between the two fifth grade classes. No-
tice that a fourth grade class (A) does work distinctly better
SUPPLEMENTING THE HILLEGAS SCALE
23
than an eighth grade class (B), and that a senior class in high
school (D) is quite clearly inferior to an eighth grade class (C).
Certainly no teacher would be satisfied to have her class doing
work typical of lower grades unless she were ignorant that they
were doing such work. Knowledge of how well a class should
do and ability to know when it was actually doing that well
would seem to be prerequisites for successfully teaching a
subject.
The purpose of the standards proposed in the following para-
graphs is merely to serve as a guide to teachers, so that they
may not be left as they have been in the past without any
definite idea of the achievements to which their classes should
aspire. Better standards will doubtless be proposed by others,
but the suggestions which follow will be worth while if they
serve merely to call forth criticism and more satisfactory
standards.
How well do " Seniors " in high school write? What quality
of compositions would you accept from " Sophomores " in high
school? These two questions were asked of advanced students
in Educational Administration. The answers were expressed
in each case by means of the list of thirty compositions employed
in the first part of this study. Without knowing the exact value
of any of the thirty compositions, thirty-six experienced super-
visors and administrators each selected two compositions from
the lot, one to represent the average or median achievement
of high school senior classes, and the other to represent just
barely passing quaHty for the same grade. Forty-six other
members of the group selected two compositions each, one to
represent the average quality obtained from sophomore classes
in high schools they had known, and the other to represent
just barely passing or 70 per cent compositions for the same high
school classes. The distribution of estimates made is shown
in Table V.
It would not be sufiicient, however, to take the estimates
of even the most able group of educators in the world as a
basis for standards of achievement. It is necessary to know
in addition just how well pupils actually write. The median
results of composition tests and measurements on the Hillegas
Scale are shown in Table VI for a number of typical school
24 TEACHERS COLLEGE BULLETIN
ESTIMATI
iS OF Quality
Sample
He Si
TABLE V
OF English Compositk
IN High Schools
Senior classes
3NS TO BE Exp:
Sophomore
Median
(8S%)
.... 2. . .
ECTED
classes
Value
8.0I
Median
(8S%)
Passing
(70%)
Passing
(70%)
8.00
Re Si
. . .3
1
7-68
Adja
. . .3
c
7.51
LuRh
. . .9
2
, 7
7.47
Eu Sw
...a
I
... 2 . .
2
7.22
6.01
Mi Su
ThBe
•••9
. . .1
6.. . ! .
13
... .2
8
5-89
ChFr
. . .4
3
8
II
5.76
ChCo
. . .4
8
I .
Q
4.98
ChDr
4
.... 2
4.97
LoCo
c
2
4.86
JoDr
4-77
CoO'T
2
2
4.36
JeMi
2
4.16
ChWi
2
4-03
RiBe
3-84
CaPe
. .... I
3.66
RoMcN....
I
1
3.31
Gr Mi '
2.81
MaRo
2.67
QuFr
2.54
Ha Da
I
2.50
PaMti
2.40
LeO'B
1-93
LeBo
1.06
Ma
1.03
Jo Si
.80
RoR
I
.68
FrKo
.004
RuDo
Number of judges 36 36 46 46
Median 7.35 5.49 7.12 5.83
systems. The lower line of this table shows the tentative
standard medians proposed by the writer as the result of this
study. Expressed briefly, the proposed standards show the
quahty of compositions to be expected from at least half of a
normal class of American boys and girls at the end of any
given school year.
Some of the medians given in Table VI are shown graphically
in Fig. 6.
It will be observed in both the table and the graph that the
SUPPLEMENTING THE HILLEGAS SCALE
25
TABLE VI
Median Achievements in English Composition by Grades
(Hillegas Scale)
School system
Median score attained in grade
IV V VI VII VIII '"^ ^^ 3rd 4th
yr. yr. yr. yr.
Lead, South Dakota^
Newark, N. J. (i school)^
Ethical Cult. Sch., N. Y. C.\
Chatham, N. J^
Salt Lake City, Utah*
Butte, Montana*
Nassau County, N. Y.'
South River, N. P
Mobile County, Alabama* . . . .
Mobile, Alabama*
Fifty -four High Schools*
Tentative Standard Medians .
357 411 4
2-39 2.51 3
4.01 4
2-95 2.85 4
3.58 3.84 4
2.34 2.80 3
2.76 3.42 3
2-31 2.55 3
3.20 3.91 4
331 3-^5 4
■H 5
or 5-
5^ 4
33 5-
72 5
39 5-
10 4
02 5.
.61 5
16 6.
■41 3
77 4-
.82 4
18 4.
.78 4
75 5-
■34 4
22
.60 4
95
57
27
74 High
29 School
37
II
56 s.oo 5.2s 5.68 5.94
62 5. 18 5.02 5.95 6.30
5.56 6.38 6.05 6.77
6.69 6.93 7.24 7.54
4.99 5.88 6.38 6.69
3.5 4.0 4.5 s.o 5.5 6.0 6.5 6.9 7.2
' From Annual Report of Superintendent of Schools, 1915-16, Lead, S. D.
'' From unpublished studies made in the department of educational administration.
Teachers College.
3 Calculated from the distributions published in Cubberley's School Organization and
Administration. World Book Company, Yonkers-on-Hudson, N. Y., 1916.
•• From Strayer's Some Problems in City School Administration, World Book Com-
pany, Yonkers-on-Hudson, N. Y., 1916.
6 From unpublished report of the Mobile School Survey by the Bureau of Municipal
Research of New York City. The Mobile County scores are for the white pupils outside
the city, while the Mobile scores are for the white pupils in the city schools of Mobile. The
seventh is the highest elementary school grade in the Alabama schools.
"From master's essay of Albion Hale I3rainard, prepared m the department of educa-
tional administration, Teachers College, 1916. High schools from over 35 different states
are included.
standard is higher than the majority of the schools have actually
achieved, although at each grade one or more school systems
have shown higher achievements than the standard suggested.
If some schools are able to meet this standard without making
particular efforts to do so, many schools should be able to meet
it when they make definite efforts to do so. It seems desirable,
at least, to set high standards " to shoot at " rather than to
make the standards conform more closely to the achievements
which the present more or less haphazard teaching efforts
secure.
There is, in addition to the above, a real value in having a
secondary standard to indicate the width of the distribution of
scores to be expected in a class. There will usually be a small
26
TEACHERS COLLEGE BULLETIN
8.00
7.00
6.00
5.00
4.00
S.OO
2.00
4
SCHOOL GRADE
7 8 9 10 11 12
^^
/
/ /
/ /
X
2^
A
/ ■'/y
%
/
""
/
/
6,
> /
s*/
/
NC-N»SiAUCa
Fig. 6
number of pupils in each class who will be unable to approach
the median standard, and for them a lower " mark " may act
as a stimulus. For this purpose, the writer suggests that 75
per cent of each class ought to write compositions of or better
than the quality indicated below:
Quality of English Composition to be Equalled or Excelled by
AT Least 75 Per Cent of the Pupils in a Class
In grade IV V VI VII VIII IX X XI XII
Quality.
30 3-5
VI
40 4.5
S.o 5.5 6.0 6.4 6.7
SUPPLEMENTING THE HILLEGAS SCALE
27
The qualities suggested above for 75 per cent of a class are
in each case just .5 of a unit lower than the qualities suggested
for 50 per cent of the class. This difference between the achieve-
ment of half and the achievement of three-fourths of a class is
practically the difference one actually finds in classes as they
are at present. The chief sources of proof for this statement
are the tables of distributions in Nassau County, New York,
where the score on each paper was practically the median of
independent ratings by three well-trained judges. It would
be somewhat unsafe to use distributions based on a single rating
of each paper by an untrained person. Taking the combined
scores from thirty-three Nassau County schools and consider-
ing the distributions of results as though each grade were a
single large class, differences were calculated by the method
indicated in Table VII.
TABLE VII
Differences Between Achievements of 50 Per Cent and of 75
Per Cent of a Grade in Nassau County
School grade
IV
V
VI
VII
VIII
IX
X
XI
XII
No. of compositions .
Median (50%) score.
75% of grade above.
1046
2.76
2.25
930
3-42
2.72
894
3-82
3.16
741
4.18
3-49
607
4-56
3-79
554
5.00
4.42
332
5-25
4.60
237
5.68
4-95
i6s
5-94
5-42
Difference (50-75).. .
•51
.70
.66
.69
• 77
•58
•6S
•73
•52
As was shown earlier in this discussion, being in the same
grade in different Nassau County schools does not mean doing
the same sort of English composition work. The above differ-
ences are larger than would be found if we were dealing with
actual classes or recitation sections. By taking all those recita-
tion sections in Nassau County containing more than 25 pupils
in the fourth grade, 20 pupils in the eighth grade, and 12 pupils
in the senior high school grade, the distributions of differences
between the achievements of 75 per cent and the achievements
of 50 per cent of the pupils in actual classes were found to be
as shown in Table VIII.
From Table VIII it will be observed that the requirement
is not too high, which suggests that the achievement of 75 per
28
TEACHERS COLLEGE BULLETIN
TABLE VIII
Frequency of Differences Between Achievements of 50 Per Cent
AND OF 75 Per Cent of Actual Classes
Size of Differences
In grade In grade In grade
IV VIII XII
. ^0—. ■?Q Hilleeas units
3 3
5 5 2
4 3 2
7 I I
2 2
3 I I
.40-. 49
. KO-. ^Q
. 60-. 60
•70-. 79
.80+
Median difference'
■59 -46 .52
' Calculated from actual cases rather than from above distributions.
cent of a class should be within .5 of a unit of the achievement
of 50 per cent of the class.
For the aid of those who may be unfamiliar with the method
of calculating the 50 per cent or the 75 per cent achievement of
a class, when the scores have been assigned to the papers merely
in terms of the nearest samples on the scale, the following table
is given.
TABLE DC
To Facilitate the Calculation of the Achievements of English Composi-
tion Classes Measured by the Nassau County Supplement to the
HiLLEGAS Scale
Sample Actual value Range of step' Length of step Sample
0
I.I
. . 0 . . . .
. . 1 . 06 . . . .
.... 0- .53 ...
54-1 -49- ■•
-53
.96
.. . . 0
. ... I.I
1-9
2.8
..1.93....
..2.81....
. . . .1.5CJ-2.36. ..
•- --2. 37-3. 32...
.87
.96
.. ..1.9
....2.8
3-8
S-O
..3.84....
..4.97....
....3.33-4.40...
.. ..4.41-5.48. ..
1.08
1.08
....3.8
....S-o
6.0
7.2
. .6.01. . . .
. .7.22. .. .
. . . .5.40-6.61. . .
. . . .6.62-7.61. . .
I -13
1. 00
....6.0
....7.2
8.0
9-0
. .8.00
..9.0 ....
7.62-8.50. ..
....8.51- ...
.89
....8.0
....9.0
' The third decimal point is omitted from these tables.
To show the method of using this table we may take the
following distribution of scores as an example, and calculate
the achievement of 50 per cent and of 75 per cent of the class.
Scored on o i.i 1.9 2.8 3.8 5.0 6.0 7.2 Total
Frequency 3 16 25 4 48
SUPPLEMENTING THE HILLEGAS SCALE
29
The median achievement will be the quality above which
there are 24 papers (48 -^ 2=24) and below which there are 24
papers. If we count the three which were rated on 1.9 and the
16 which were rated on 2.8 we will have only 19 papers, and in
order to reach the middle of the distribution we shall need to
count 5 of the 25 papers rated on 3.8. We may assume that
the 25 papers are spread evenly over step 3.8, extending (see
Table IX) from ^.;^t, up to 4.40 at equal intervals. If we count
out 5 of the 25 we shall pass over one-fifth of this step which
is 1.08 units long; in other words, we shall have gone upward
from T,.^T, a distance of .22 of a unit (1.084- 5=. 2 2) and we shall
therefore have arrived at quality 3.55, which is the median
quality of the distribution.
In order to find the achievement made by 75 per cent of the
class, we shall need to count only one-fourth of the way into
the distribution, — that is, we shall need to count out the lowest
12 pupils (48-^-4=I2). After counting the 3 papers scored on
1.9 it will be necessary to count out 9 (12 — 3=9) of the 16
scored on 2.8. Assuming that the quaHties of the papers are
distributed evenly over step 2.8 from quality 2.37 to quality
3.32, we shall count out nine-sixteenths or .56 of the step, which
is .96 of a unit long. This will take us to a point .54 of a unit
(.96X.56=.54) above quality 2.37, which will be quality 2.91,
the achievement made by 75 per cent of the class.
The table for the original Hillegas Scale which corresponds
in its use to Table IX is given here as Table X.
TABLE X
To Facilitate the Calculation of Class Achievements on the
Hillegas Scale for English Composition
Step No. Value of sample Range of step Length of step Step No.
o 0.00- .91 91.
1.83 92-2.21 I -SO-
2 2.60 2. 22-3.14 93 2
3 3 69 3. 15-4. 21 1.07 3
4 4-74 4.22-5.29 1.08 4
5 5 85 5-3^6.30 1. 00 S
6 6.75 6.30-7.23 93 6
7 7 72 7.24-8.05 81 7
0 8.38 8.05-8.87 82 8
8 9.37 8.88- 9
30
TEACHERS COLLEGE BULLETIN
It is not proposed that any teacher should use the Hillegas
Scale in any form as the basis for all her marks in EngUsh com-
position. The scale is a measuring instrument and not a peda-
gogical device. In her instruction a teacher may at one period
be laying stress on punctuation and give her class marks accord-
ing to their success in punctuation. At another period the
teacher may be placing her emphasis upon the choice of words,
and the class marks may well reflect this emphasis. But at
some two or three periods during the year teachers should meas-
ure the general results that are being obtained by this instruc-
tion. For this purpose a number of teachers should cooperate
with each other in rating papers on. some form of the Hillegas
Scale. When such practices become more general, we may hope
that results in English composition writing will become uni-
formly better.
APPENDIX
More than five years have passed since the above discussion
was first published. During this period a considerable amount
of real progress has been made in the science of measuring the
quality of English compositions. Professor Earl Hudelson^of
the University of West Virginia has been one of the most fruit-
ful workers in this field.^
He has found, for example, that pupils write compositions of
a much lower quality when they use the " What I Should Like
To Do Next Saturday " topic than when they write on such a
topic as " My First Lie." The average results obtained from
" My First Lie " are one full unit better than the average results
from the topic on which pupils in Nassau County wrote. If
an eighth grade teacher secures from her pupils compositions
on the shorter topic, she should subtract i.oo from their median
score before comparing it with the records made by pupils on
the " Next Saturday " topic. Such facts indicate the necessity
of having a uniform topic or set of topics for testing ability
in English composition, and the importance of following uniform
directions.
Directions for Obtaining Compositions
Have each pupil clear his desk and provide himself with pen,
ink, a sheet of the usual composition paper and a blotter. The
pupils should be asked to write (at the top of the sheet) their
names, ages (at last birthdays), and school grades. If papers
from more than one room or school are to be scored at the same
time, it would be well to obtain from each pupil before the test
begins a written record that will fully identify his paper. While
the pupils are writing their names and other desired informa-
tion, the examiner should write one of the four following topics
on the blackboard. Be sure to write plainly and where all
can see.
The Funniest Thing I Ever Saw
The Saddest Event of My Life
The Happiest Hour of My Life
The Most Unpleasant Hour of My Life
'Hudelson, Earl, Aims, Methods and Measurements in English Composition.
32 TEACHERS COLLEGE BULLETIN
Call attention of the pupils to the topic in the following words :
" I have written on the blackboard the topic for a story. You
are to write a story about the [topic]. You may make up a
story if you need to, but most of you can tell a real experience.
You may have twenty minutes in which to write. See how
interesting a story you can tell."
All papers should be collected at the end of twenty minutes,
whether the pupils have finished writing or not. However, in
assigning scores to the papers, nothing should be subtracted from
any pupil's mark because of an incomplete story. Each paper
should be marked on the quality of the writing actually done.
Compare the general quality of the pupil's composition with
the general qualities of the various samples on the scale. Assign
to each composition the numerical value of that sample on the
scale which most nearly equals it in general merit. Go up the
scale until you reach compositions that are distinctly better than
the one being rated ; go down the scale until you find composi-
tions that are distinctly poorer than the one being rated ; and
then compromise on some value between these two limits. In
order to secure accuracy, it would be wise to have each compo-
sition rated on the scale by at least three competent judges, each
making his rating without knowledge of what the others have
assigned. The average or the middle rating of the three judges
may be taken as the one to represent the final value of the
composition.
Directions for Improving the Judgments of Teachers
In order to improve the abilities of teachers to judge the gen-
eral qualities of English compositions, the other twenty-one
compositions (rated by the 139 judges but not included in the
Nassau County Supplement) are reprinted here as practice ma-
terial. Let each judge read one of these compositions, compare
it with the samples on the scale, decide carefully the numerical
value that seems appropriate to it, and then turn to the Key on
the last page of this booklet and compare his judgment with
the value quoted there.
One should not be satisfied to discover that he has over-
estimated or under-estimated a composition, but he should care-
SUPPLEMENTING THE HILLEGAS SCALE 33
fully re-examine the composition and the scale to see why the
median of 139 judgments gives it the value that appears in
the Key. After discovering why one has made an error, he may
try again on another composition, preferably not the one imme-
diately following, because of the danger that he might remember
its value from the Key.
A careful rating of ten or twelve of these practice composi-
tions, followed in each case by a comparison with the Key and
a thorough re-examination of the scale to see why one's ratings
differ from those in the Key, should reduce one's " personal
equation " considerably. One may consider himself a reasonably
competent judge when the average difference between his ratings
and the true values has been reduced to less than .5 of a unit.
The true values of the following compositions are given in
the Key on page 38.
Ri Be 4.03
We have all ready secured bases and other things to have a good game of base-
ball and if we finish the game in time we will go into the woods and make a hute
and have a kind of party. We are going to have cake and crakers. We hope to
have a good time for we are going to have games of all sorts ,
Ch Co 5.76
My desire for enjoyment next Saturday would be a camping out party where
I could obtain plenty of fishing, hunting and water sports, of which I am fond of.
I would like to have my cousins accompany me, as they know a great deal
about fishing, and know how to track wild game, owing to their previous expedi-
tions which they made with their uncle last summer.
It seems to me that our journey would be all the more pleasanter if we traveled
by canoe, and make one or two portages when possible.
I think this would be a pleasant way to spend next Saturday, as nothing is
more injoying than to be out in the open country, where you can breath the fresh
air of the woods and feel at home once more.
Ha Da 2.54
I intend to go next Saturday is to sleigh riding, and part of the time to go ice
skating if there is any ice skating, do my lessons, do my other work, and help
my mother^ help my brother, and sister.
And do my poia lessons, and take my ponia lesson, and go down celler and
chop wood for my mother, go the to the store for errands for my mother, and
wipe the dishes, and bring up coals, wood for my mother, if I can make a kite
and when it dries I will fly it for a little while and come in if it get cold and get
warm and go out for a nother Httle and do the same thing if there is any more
thing to do I wiU do it.
Jo Dr 4.86
Next Saturday I should like to Play ball in the morning till about 10 o'cloack.
Then I should like to go to the beach and go in for a swim. In the Afternoon
I should like to go to the city and see the ball game between the giants and Phila-
delphia. In the evening I would like to go to the the theatre.
^4 TEACHERS COLLEGE BULLETIN
Ch Dr 4.98
If the weather be clear, my intentions are all planned out for next Saturday.
In the morning I have made a promise to pitch, (for a team just selected of
fairly good players), who are to play the " Happy Nine " of Bullshead. The
game is called at nine o'clock and the players are to be there about eight, so as
to give us a chance to get warmed up (as base ball players calle it).
I am a pitcher that can not be found every day of the year. As I have played
in a great many games of base ball. I have been practicing all this week after
school hours. And think I can do the trick.
Ch Fr 5.89
Next Saturday I should like to have a good game of baseball because I like that
game very much.
I would not care to win a game by unfair means nor by a large score but some-
thing like one to two or one to nothing in our favor with every man watching his
own game and playing to the best of his ability.
I IDie baseball because you have to be quick to play a good game and keep
your head when you are playing against odds or when your are in a tight place —
for instance You are two runs ahead of your opponent and they have two men
on base and a good batter up to bat in the last half of the nineth inning, — then
a good many pitchers and players get discouraged thinking there are no chances
to win, but the ones that stick to it until the last man is out is the man or team
for me.
Qu Fr 2.67
I am going to play ball and help my mother and I am going riding, deliver papers
then with my money go by a new ball.
then I will feed my rabbits and chickins after they are feed I am going to Scouts
an practice
then when I come home I'll go play.
At night go to the movies after movies go get a nickels worth candy
Ad Ja 7.68 What I should like to do when I leave High School
My highest ambition after I leave high school is to enter coUege and successfully
complete my course there and then enter upon a career of service for others.
From a child I have always longed to be among the poor and lowly not at first
for the purpose of doing missionary work but because my life had had so much
happiness that I can not bear to think of anyone being unhappy.
I have always had a great admiration for missionaries but their lives seemed
to be so set apart, so far above anything I could become or hope to become. Not
until a few months ago did the idea come to me with any great force but from
that time on I was determined to become a missionary not only that I may share-
my happiness with others, but to teach them of the loving God, believing that
through Him, they may be brought into lives of happiness and usefulness.
Fr Ko .68
What I should do Saturday play boll would The flower hil in the afftermoon and
haue A chatch would the chatch on the boll lot I am going to patch on the
manhasset team We wone the gam 16 to 4 and thun we went home
Ro McN 3.66
Next Saturday being that I have an holiday, I expect to go to Washington
D.C. to spend Saturday and Sunday looking at the Pension Building, Libraies
and the Captail etc. Into which I am very much interested in.
I would like to see the large streets, and especially visit the cabinet the large
room and hear the statesmen talking
SUPPLEMENTING THE HILLEGAS SCALE
35
Gr Mi 3.31
I should like to-do next Saturday to help my mother and clean all the house and
sweep the floors and scrob. and wash windows and chop wood for her and carrie
it in the house and obey her and my father. Go to school hight on monday morn-
ing and to store and go out shopping with her. And if it is early mother and I
would go to a show. Then I should help my big sister and clean house for her
and wash windows and sweep and shake all the carpets and make the beds and
open all the windows. Then mother and my sister in-law would give me fifty
cent a piece.
Je Mi 4.36
I should like to go a way for a long journey to California. And see the nice
fruit trees and see how large they are. It is Very nice out there. For I have friends
out there and they say it is very nice?
There is large trees out there that they have made tunnels through and places
where horses can go through
There is lots of apple, and Oranges in California They send a lot of fruit to
New York. I an in Roslyn now and do not like it Much? I should like to stay
there for three months but I think only can stay for a few Weeks. But all the
same I like to go and see my friends?
Pa Mu 2.50
What I shall do Saturday
I shall go out to my aunt in Linderhurst
When I get out their, I will go out to the chickens and then see the two dogs
When uncle otto comes home from worke Saturday night. Wee will
get up early in the morning
and get the chickencoop
clean
And in the might I will go to the move picturs with uncle otto
Le O'B 2.40
I am going to school tomorrow. Are you going to play with me.
I am going to the theatre to night. Are you going to night to. I am going home
If you will play with me I will go. Alright are you coming now to play
Don't forget to come. " Yes, I won't forget to come. I am going to play school
at home. I will go with you I won't with you. Are you playing with us No
I ant playing with you.
I Am going to play bass ball. Are you going to play to. " Yes, I am going
to play.
Co O'T 4.77
Next Saturday a number of girls in my class are going on a tramping trip,
sometimes called a Nature Study trip. As a very merry and jolly bunch of girls
are going, I should also like to.
And if I am not allowed to go on this Trip, I am making reckoning on spend-
ing The day with my Aunt, For in The evening. Some friends of hers are planning
to bring a surprise party for her.
But as it is I really do think I would rather go on The trip, For I am sure I
would have more fun.
RoR .80
I intend to go slealy rady next Saturday and mach a baby and pout a lout of
chilon on.
And aftir we done we will go home and mach a big fout and mach two and we
will have a snou ball find and we will rush om the eneany and we will catep there
fout and we will traif thin home and we will ceap ther fout fou a price and there
will mach a nether and we will fout thin again and we will have a good time
36
TEACHERS COLLEGE BULLETIN
Lu Rh 7.51 What I should ^ike to do when I leave High School
When I am graduated from Oceanside High School, I should like to study
Chemical Engineering in the Carnegie Institute of Technology at Pittsburgh,
Pa. I should like to finish my work at college as soon as possible in order to begin
the more serious business of life. This work has always held for me considerable
fascination, and I am deeply interested in it. Material wealth and riches need
not be the result of my choice of occupation, but rather satisfaction with my
work, an interest in it and, above all, the realization that my life has not been
wasted.
Ec Si 8.01
Saturday I should like to transport myself to one of inumerable glens dotting
the Southern portion of our country. It would be a shady, cool retreat, where
moss covered rocks form natural seats, — where a crystal brook would babble
and murmur of the wood creatures, — of the plants, and of the spring, as it winds
its way among the monarchs of the woods, — where birds would serenade me with
their sweet, warbling melody. And there I should study the wonderful manifes-
tations of the power of God, — the bubbles of snowy clouds floating across the
infinite surface of the azure sky, — the animmals and their homes. And then, per-
haps I should roam thro the leafy aisles of the wood and study the inumerable
flowers and plants, and breathe nature-given balm — their perfume. The colorful
pictures, — the sky with its exquisite tints, — the flowers, — the birds, — the brook, —
would give me more pleasure and information than the greatest pictures ever
painted. Such is my dream of a perfect day, — and to sleep beneath the gleam-
ing stars set in the soft, dark velvet of the sky, and breathe the invigorating airs
of the woods, — its perfect ending.
Jo Si i.oj
1 ever af ten noon Three olock want I go home I took my gulf club and play. And
I had lot of fun and sum Time I play bast ball in the night. And The Saturday
I go to the gulf, club want I go there we bast ball and. I make lot of money and
in Sunday
Eu Sw 7.47 My future plans
As a general thing a boy of fifteen or sixteen has no definite plans for the
future. His ambitions are extremely eccentric and in most cases absurd. I believe
that one of youth's first ambitions is to become an actor. He attends some theatre
or perhaps the circus and is deeply impressed by blase of everything. This passion
usually last about two years. A boy's future depends upon his temperament
and the impressions dictated by in the ordinary course of life. I am glad to say
that I have emerged from the previously mentioned age and after careful con-
sideration have come to the conclusion that a course at either West Point or An-
napolis is the zenith of my ambitions. On the whole I prefer Annapolis but if_ by
any chance I should be unable to secure an appointment there I would not object
to West Point. Both Aacdemies offer the widest opportunities to a young man
both from an educational and fundamental standpoint. I do not believe that a
young man could at the present time chose a more honorable or opportune future.
Ch Wi 416
I am going to playball, in a corner lot near are house. I might go to Great
Neck in the afternoon if it does not rain. I would like to go on a trip to New
York. I think I can go. I would like to go in the woods and pick May flowers
and Violets if they are out. I should like to go and see a ballgame, and a good
one too. I might can go to the railroad station and see if it is a package there.
I would like to see the Great Neck Scouts play ball tomorrow.
SUPPLEMENTING THE HILLEGAS SCALE
Additional Data on Standard Scores
37
The younger pupils in each grade will usually write better
compositions than the older pupils, because they have somewhat
greater academic ability than the older pupils. All results from
academic or intelligence tests should therefore be tabulated by
age within each grade} If a class happens to be composed of
bright young children, the tentative standards given in the original
discussion (page 19) will be too low; while if the class is com-
posed of retarded older pupils, those standards will be found
too high. The writer recommends, therefore, that only the
normal-age group in each grade be used for comparing one school
with another. The six-year-old pupils are to be considered
normal for the first grade, the seven-year-olds for the second
grade, and so on through all the grades. As a matter of actual
fact it often takes more ability than an average pupil possesses
to be in the eighth grade of the usual school at age fourteen, but
theoretically fourteen-year-olds are of normal age in the eighth
grade.
The statement of median scores given below includes, there-
fore, not merely the tentative standards proposed by the writer
five years ago, but also the average of the actual qualities of all
compositions measured in each grade, and the median achieve-
ments of pupils of normal age in each grade. Older pupils should
not be expected to do as well as, and younger pupils should
be expected to do better than, the normal-age group in each
grade.
Median Achievement in English Composition by Grades
Grade IV V VI VII VIII IX X XI XII
Standard (Tentative) 3.5 4.o 4-5 5-° 5-S 6.0 6.5 6.9 7.2
Actual Average 3-° 3-6 42 4-7 5-3 5-2 5-9 6.3 6.7
Normal Age Pupils 3.0 3.6 4.1 4.6 5.1 5-5 5-9 6.3 6.7
The above scores represent the work of each grade at approxi-
mately the middle of the school year. Some allowance would
'For evidence on this point see the writer's report on tests given in St. Paul
(Minn.) School Survey and in the Baltimore (Md.) School Survey.
38
TEACHERS COLLEGE BULLETIN
need to be made in each grade for results taken at the beginning
or at the end of the year.
Sample
Ri Be
Value
4,0?
ChCo
Ha Da
5
2
76
5/|
To Dr
4
86
ChDr
ChFr . ..
4
c
98
89
67
QuFr
2
Key to Compositions on Pages 33 to 36
Value
7.68
Sample
Ad Ja
FrKo 68
Ro Mc 3.66
GrMi 3.31
JeMi 4.36
Pa Mu 2. 50
Le O'B 2.40
Sample Value
CoO'T 4.77
RoR 80
Lu Rh 7.51
He Si 8.01
Jo Si 1.03
Eu Sw 7.47
Ch Wi 4. 16
LIBRARY OF CONGRESS
0 019 840 738 0
Live Music
Archive
Librivox
Free Audio
Metropolitan Museum
Cleveland
Museum of Art
Internet
Arcade
Console Living Room
Open Library
American
Libraries
TV News
Understanding
9/11